Title: Relational fuzzy clustering to identify non-linear structures
Authors: Maria Brigida Ferraro - Sapienza University of Rome (Italy) [presenting]
Paolo Giordani - Sapienza University of Rome (Italy)
Abstract: Classical (hard or fuzzy) algorithms usually detect clusters by computing the Euclidean distance among pairs of objects. They are based on the linearity assumption and, therefore, do not identify properly clusters characterized by non-linear structures. In order to overcome this limitation, the so-called geodesic distance can be considered, where the linearity assumption holds locally. In fact, the geodesic distance between two neighboring objects is equal to the Euclidean one whilst, in case of two faraway objects, it is equal to the shortest path in the graph connecting them. The aim is to propose some relational fuzzy clustering methods for non-linear data.